{
  "id": 28571,
  "title": "Helping AI models to meet the real world",
  "url": "https://urgent.news/2026/07/14/helping-ai-models-to-meet-the-real-world-28571",
  "topic": "ai",
  "section": "AI",
  "published": "2026-07-14T20:25:00.000Z",
  "source": {
    "name": "MIT News Research",
    "slug": "mit-news-research",
    "url": "https://news.mit.edu/2026/helping-ai-models-meet-real-world-0714"
  },
  "original_language": "en",
  "account": "In recent years, businesses have increasingly adopted artificial intelligence (AI) to improve forecasting, planning, and decision-making processes. However, many of these AI systems lack the specific, detailed information about the organization itself, limiting their effectiveness. Devavrat Shah, a principal investigator at MIT's Laboratory for Information and Decision Systems and a professor in Electrical Engineering and Computer Science, has been working on designing AI methods that can handle real-time decision-making using limited computational resources. Shah explains that his goal is to develop methods that can extract information from data at scale in an effective manner. He has co-founded a spinoff company called Ikigai Labs, which has developed a foundation model for tabular, time series data based on research in his lab. This model can take input from enterprise data in various formats, continuously and at scale, learning as it goes along by testing predictions against real outcomes. Unlike most AI models that use text and images as input, this system uses structured tabular data, such as the row-and-column format found in spreadsheets. This allows for real-time planning on a vastly larger scale. Ikigai's technology is particularly useful for large businesses like consumer goods manufacturers and pharmaceutical companies. For example, a consumer electronics company could use the system to predict product sales in different regions, optimize pricing and promotions, and improve overall business operations by making interdependent decisions over time. Shah's model has been acquired by Celonis, a company that specializes in digitizing and automating operations for over 1,400 large companies worldwide. As the chief scientist at Celonis, Shah hopes Ikigai's software can integrate with these companies' data and business processes to provide detailed models for simulation, prediction, and decision-making. By focusing on structured or time-domain data, Shah believes Ikigai offers a cost-effective version of AI that is broad enough to be highly valuable.",
  "summary": "Through research and entrepreneurship, Professor Devavrat Shah is helping to design methods that can handle constant decision-making using limited computational resources.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "MIT News AI",
        "title": "Helping AI models to meet the real world",
        "url": "https://urgent.news/2026/07/14/helping-ai-models-to-meet-the-real-world",
        "published": "2026-07-14T20:25:00.000Z"
      }
    ]
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}